Using the per-revision bisect providing the bisect results,
Good build:55.0.2844.0(Revision: 415023).
Bad build: 55.0.2845.0: (Revision:415522).

You are probably looking for a change made after 415439 (known good), but no later than 415440 (first known bad).

CHANGELOG URL:

The script might not always return single CL as suspect as some perf builds might get missing due to failure.

https://chromium.googlesource.com/chromium/src/+log/540399d5f34abfcae1828e2c36835ffae391e319..2ef7df4c03595db39c1b0c303ec8a822f0bf30c9

From the CL above, assigning the issue to the concern owner 

@trchen - Could you please check whether this is caused with respect to your change, if not please help us in assigning it to the right owner.

Review-Url: https://codereview.chromium.org/2295483003

Note: Adding RB-Stable label since this is a recent regression, please feel free to edit it if this not the case.

Thanks !

Just to update, still able to reproduce the issue on Windows 10, MAC 10.11.6 for chrome version 56.0.2900.0. Issue is marked with a blocker label and M55 is already in Beta.

@trchen: request you to please take a look into it and provide an update on it.

Thanks.!

This is a M55 regression, it should block stable for M55, not M56.

**** Bulk edit -  please ignore if not applicable ****

A friendly reminder that M55 Stable is launch is coming soon! Your bug is labelled as Stable ReleaseBlock, pls make sure to land the fix and get it merged into the release branch ASAP so it gets enough baking time in Beta (before Stable promotion). Thank you!

Will not fix. Working as intended.

Their "picture book" effect abused a compositing bug previously in WebKit/Blink. It worked in Firefox because their framework applied different CSS to workaround the correct behavior in Firefox.

Full story: (you may TL;DR)

Overflow clip should not affect elements that are not positioned by it. For example:
<div style="position:relative; overflow:hidden">
    <div style="position:fixed"></div>
</div>
The fixed-pos element will ignore the clip. Prior to my CL, our compositing code path made a mistake not to escape clip correctly. (The software path wasn't affected by this issue.)

To achieve the intended behavior, CSS clip can be used instead. Unlike overflow clip, CSS clip affect all descendants. So something like this can be done:
<div id="g-map-sandstorms">
  <div style="position:absolute; left:0; top:0; width:100%; height:100%; clip:rect(auto,auto,auto,auto);">
    <div class="g-text-block">...</div>
    <div class="g-media g-ai2html-poster">...</div>
  </div>
</div>

I've reached out to techfeedback@nytimes.com . Let me know if we've worked with them before and know a contact person.

Just to to update the latest behavior of the bug, Issue is still observed on chrome latest Canary M57-57.0.2926.0

trchen @ Could you please let us know is there any recent update available on this issue?

Thanks!

Didn't get a reply from them. Nothing we can do if they don't fix their own bug.